|
Message
From: Matjaz Breskvar<phoenix@o...>
Date: Thu Jun 30 13:11:07 CEST 2005
Subject: [openrisc] [or1ksim #107] Remove useless calls to m{f,t}spr
* Gy?rgy 'nog' Jeney (nog@s...) wrote:
> > * Gy?rgy 'nog' Jeney (nog@s...) wrote:
> > > Hi,
> > >
> > > There has been a nice amount of traffic on openrisc in the last days, may aswell
> > > add my own :)
> > >
> > > ChangeLog:
> > > * Remove m{f,t}spr calls where we can access the spr directly.
> >
> > i would think that it is better (for readability) to have
> > m{f,t}spr_internal function, that is an inline of what you suggest. it's
> > just as fast and you still keep the m{f,t}spr abstraction...
>
> mfspr_internal(spr)
> cpu_state.sprs[spr]
>
> What's the difference for readability? The one has round brackets, the other
> square ones and they're even the same length. What abstraction are you thinking
> of? The whole idea is that there should not be any abstraction. As soon as we
in the openrisc this is implemented as a bus. ie, it can have more than one
destination. in this case you'd need to do more ten one cpu_state.sprs[spr]
for certain spr while just one mfspr_internal(spr) ... for now this is not
an issue AFAIK, so consider this as an info,... i leave it up to you to
decide if it should be commited.
best regards,
p.
|
 |
Browse
